Idea

Dead code are pieces of code that do not affect the output of a program. The removal of these unnecessary code lines can be defined as Dead Code Elimination. These code lines do not serve any purpose in the main program flow because these codes will never be executed mainly because the condition for their execution is logically impossible/infeasible. This removal results in performance enhancements.

Background

Dead Code Elimination is an optimization that removes code which does not affect the program results. You might wonder why someone would write this type of source code, but it can easily creep into large, long-lived programs even at the source code level. Removing such code has several benefits: it shrinks program size and it allows the running program to avoid executing irrelevant operations, which reduces its running time. It can also enable further optimizations by simplifying program structure.

For example, consider the following code:

foo <- function() {
  a <- 24
  if (a > 25) {
    return(25)
    a <- 25 # dead code
  }
  return(a)
  b <- 24 # dead code
  return(b) # dead code
}

In functions, after calling return, the following code would not be executed, so it is dead code and can be eliminated. In this example, resulting in:

foo <- function() {
  a <- 24
  if (a > 25) {
    return(25)
  }
  return(a)
}

Also, after constant propagating and folding we would get:

foo <- function() {
  a <- 24
  if (FALSE) { # dead code
    return(25) # dead code
  } # dead code
  return(a)
}

So it could be reduced to:

foo <- function() {
  a <- 24
  return(a)
}

This dead code optimizer also removes code after next or break calls.

Example

Consider the following example:

code <- paste(
  "i <- 0",
  "n <- 1000",
  "while (i < n) {",
  "  if (TRUE) {",
  "    i <- i + 1",
  "  } else {",
  "    i <- i - 1",
  "  }",
  "}",
  sep = "\n"
)
cat(code)

## i <- 0
## n <- 1000
## while (i < n) {
##   if (TRUE) {
##     i <- i + 1
##   } else {
##     i <- i - 1
##   }
## }

Then, the automatically optimized code would be:

opt_code <- opt_dead_code(list(code))
cat(opt_code$codes[[1]])

## i <- 0
## n <- 1000
## while (i < n) {
##   i <- i + 1
## }

And if we measure the execution time of each one, and the speed-up:

bmark_res <- microbenchmark({
  eval(parse(text = code))
}, {
  eval(parse(text = opt_code))
})
autoplot(bmark_res)

speed_up(bmark_res)

##            Min.  1st Qu.   Median     Mean  3rd Qu.     Max.
## Expr_2 139.6213 130.6668 101.3909 97.91606 80.83488 85.56513

Implementation

The opt_dead_code optimizer performs two main tasks: